280 likes | 448 Views
Proposal of Fast AP Discovery. Authors:. Date: 2012 -01- 11. Abstract. This document proposes faster AP discovery which doesn’t increase packet congestion so much . Conformance w/ TGai PAR & 5C. Reviews of “Fast AP Discovery” discussions in Atlanta.
E N D
Proposal of Fast AP Discovery Authors: • Date:2012-01-11 Katsuo Yunoki, KDDI R&D Laboratories
Abstract • This document proposes faster AP discovery which doesn’t increase packet congestionso much. Katsuo Yunoki, KDDI R&D Laboratories
Conformance w/ TGai PAR & 5C Katsuo Yunoki, KDDI R&D Laboratories
Reviews of “Fast AP Discovery” discussionsin Atlanta • Active Scanning by Probe Request/Response may occupy much air-time to harm primary purposes of WLAN, which are associations and data communications, as the smart-phone market grows. • A STA can’t transmit any packets before receiving a packet from the master device (AP)in radar-coexisting bands. • Passive Scanning by detecting Beacons takes some 100 msec to discover AP. It was commented in some contributions thatfaster AP discovery was preferable. Katsuo Yunoki, KDDI R&D Laboratories
Is Active Scan really fast? (1/2) • A STA isn’t always in an AP area. In that case, Active Scan is just battery consuming for mobile devices. • When a STA enters an AP area, Active Scan may be useful for discovering that APin fast. But the STA doesn’t know when should start Active Scan. Probe Request STA STA Probe Request AP Battery consuming!! When starts Active Scanning?! Katsuo Yunoki, KDDI R&D Laboratories
Is Active Scan really fast? (2/2) • Currently, a STA transmits Probe Request frames every some 10 seconds for less battery consumption. • So it takes 10 seconds in the worst case for Active Scan to discover an AP. (We can say that average time will be 5 seconds.) • Do you think it’s fast? Katsuo Yunoki, KDDI R&D Laboratories
Negative effects by Active Scan • AP discovery by Active Scan isn’t so fast. • Air-time occupancy will be proportionally increased to the number of STAs. It is needed to consider a large volume of STAs in FILS use cases. • So... This submission proposes not to use Active Scan in FILS. Katsuo Yunoki, KDDI R&D Laboratories
Effective cases of Active scanning • When the STA was connected with a specified AP, • the STA knows the destination MAC address and SSID of the AP. Probe Request should be by unicastand with a specific SSID (not wildcard-SSID). • When the STA starts data communication as internet browser initiated on the device (ex. @ train station), • Probe Request may be by broadcast with wildcard-SSID. But it should be just one shot transmission. • It is desirable for preventingunnecessary Probe Responses to send Probe Request with specified SSIDs. Katsuo Yunoki, KDDI R&D Laboratories
Effective cases of Active scanning(cont.) • 3. When the STA searches stealth APs, • the STA knows SSID of the APs. Probe Request should be with a specific SSID. Katsuo Yunoki, KDDI R&D Laboratories
Reminding of proposals in doc. 1414r2(F2F in Atlanta) • TGai compliant STAs shall not use the wildcard SSID in SSID element of “Probe Request”. • TGai compliant APs shall not transmit “Probe Response” for “Probe Request” with the wildcard SSID. FILSfeatures shouldn’t be provided to STAs sending Probe Requests with wildcard SSID. It serves more effectiveness on FILS and will bring clearer circumstances. Katsuo Yunoki, KDDI R&D Laboratories
Faster Passive Scanning • Faster Passive Scanning may be achievable by shortening the interval of Beacons. • But air-time occupancy also will be increased. • This submission proposes smaller sized Beacon called Sub-Beacon just for AP discovery. Katsuo Yunoki, KDDI R&D Laboratories
Proposal-1Sub-Beacon • Sub-Beacon consists of SSID only. • It is transmitted in shorter duration than the regular Beacon. • When a STA finds Sub-Beacon, it waits for the regular Beacon for compensating information lack. Regular Beacon ex. Sub-Beacon Time Beacon interval / 3 Katsuo Yunoki, KDDI R&D Laboratories
Searching Process • A STA has to monitor all channels one by one for detecting Beacon. • When Sub-Beacon introduced, monitoring duration for each channel may be shortened. So Passive scanning will be completed in short required time. • For detecting the legacy APs, the second round of AP search might be executed by regular (non-FILS) monitoring durations for each channel. Katsuo Yunoki, KDDI R&D Laboratories
Graphics of proposal • Conventional method • Proposed Passive Scan Monitoring window Monitoring window shorter ! CH1 CH1 CH2 CH2 CH3 CH3 • Monitoring window for each channel will be shortened by introduction of Sub-Beacon. • Completion time of Passive Scan will be shortened. • AP discovery by Passive Scan will be faster. Katsuo Yunoki, KDDI R&D Laboratories
Composition of Sub-Beacon • Sub-Beacon is transmitted in Action frame. Destination address is broadcast. • Data rate can be higher (ex. 11Mbps). It’s not necessary to consider backward compatibility for legacy STAs. Sub-Beacon (Action Frame) Katsuo Yunoki, KDDI R&D Laboratories
Air-time occupancy (1/2)Data size (Assumption) Regular Beacon (*) Sub-Beacon (Action Frame) (*) More fields may be added for Regular Beacon. Katsuo Yunoki, KDDI R&D Laboratories
Air-time occupancy (2/2) Occupied Time = (DIFS + CW + aPreambleLength + aPLCPHeaderLength/DATARATE) + (Bytes * 8/DATARATE) • aSlotTime: 20us • aSIFSTime: 10us • aPreambleLength: 144us • aPLCPHeaderLength: 48bits • aCWmin: 31 • aCWmax: 1023 • DIFS: 50us • CW: 310us Regular Beacon: (50+310+144+48/1.0) + (108 * 8 / 1.0) = 1,416 (usec) Sub-Beacon: (50+310+144+48/11) + (19 * 8 / 11) = 522 (usec) • Air-time occupancy by Sub-Beacon is about one third of the regular Beacon. • Introducing Sub-Beacon is more effective than shortening the interval of regular Beacon. Katsuo Yunoki, KDDI R&D Laboratories
Proposal-2Operating Channels IE for Beacon • Operating Channels IE is also proposed for Fast AP discovery when the AP operates in multi-bands. ElementID Length Operating Channels Octets: 1 1 5 Reserved Operating Channel in 5735-5835MHz Operating Channel in 5170-5330MHz Operating Channel in 2.4GHz Operating Channel in 5490-5710MHz Bits: 2 5 11 8 14 Katsuo Yunoki, KDDI R&D Laboratories
Composition of Operating Channels IE • Operating Channel bits are set as true. • 2.4GHz (=14bits) • 5170-5330MHz (=8bits) Katsuo Yunoki, KDDI R&D Laboratories
Composition of Operating Channels IE(continued) • 5490-5710MHz (=11bits) • 5735-5835MHz (=5bits) Katsuo Yunoki, KDDI R&D Laboratories
Searching Process for 5GHz bands • When Operating Channels IE in the 2.4GHz bandreceived, the STA can go to the channels in the 5GHz bands directly and find Beacons in those channels in faster. Katsuo Yunoki, KDDI R&D Laboratories
Notes • 60GHz and Sub-1GHz bands are not considered for Operating Channels IE. Because the assumed coverage areas might be totally different from ones of 2.4/5GHz bands. Katsuo Yunoki, KDDI R&D Laboratories
Summary • This submission proposes: • Limitation of Active Scanning. • Sub-Beacon for faster Passive Scanning. • Operating Channels IE for fast redirection to the 5GHz channels. Katsuo Yunoki, KDDI R&D Laboratories
Questions and Comments Katsuo Yunoki, KDDI R&D Laboratories
Straw Poll - 1 • Probe Request with wildcard SSID shall be prohibited at FILS compliant STAs. • Probe Response shall not be transmitted against Probe Request with wildcard SSID at FILS compliant APs. • Y/N/A = Katsuo Yunoki, KDDI R&D Laboratories
Straw Poll - 2 • Sub-Beacon for faster Passive Scanning should be considered. • Y/N/A = Katsuo Yunoki, KDDI R&D Laboratories
Straw Poll - 3 • Operating Channels IE should be considered for faster redirection to 5GHz bands. • Y/N/A = Katsuo Yunoki, KDDI R&D Laboratories
References • doc.: IEEE 11-11/1031r0 (Root) • doc.: IEEE 11-11/1064r0 (Vodafone, Nokia) • doc.: IEEE 11-11/1413r1 (KDDI R&D Laboratories) • doc.: IEEE 11-11/1521r1 (FOKUS) • doc.: IEEE 11-11/1523r3 (Qualcomm) • doc.: IEEE 11-11/1548r1 (Marvell) • doc.: IEEE 11-11/1559r0 (Nokia) Katsuo Yunoki, KDDI R&D Laboratories